Rotex Technologies Releases Nuclear-SQEP: The UK's AI Model for Nuclear Engineering Specialists
May 2026 · Rotex Technologies
Rotex Technologies Ltd has released Nuclear-SQEP — the first open-weight artificial intelligence model purpose-built for the nuclear engineering sector — marking a defining milestone in the UK's journey to harness AI within one of its most safety-critical and knowledge-intensive industries. The model is now publicly accessible via Hugging Face and represents the world's first domain-trained AI with deep competency in nuclear SQEP frameworks, regulatory standards, engineering knowledge management, and decommissioning intelligence.
Rotex Technologies is a UK-based engineering design and consultancy company with over 25 years of combined experience delivering advanced engineering solutions across nuclear, renewable energy, oil and gas, and advanced manufacturing sectors. The company has an established track record in nuclear supply chain work — most recently collaborating with Sellafield Ltd on a VR-enabled manufacturing feasibility study for the Intermediate Level Waste container Final Assembly Line concept, delivered under the LINC with Sellafield Ltd scheme.
Building on that engineering foundation, Rotex Technologies has now entered a new domain: the development of specialist AI models for regulated industries. Nuclear-SQEP is the first output of this programme, and it is offered to the wider nuclear community as both a demonstration of capability and an invitation for collaboration.
A Critical Challenge for the UK Nuclear Sector
The concept of a Suitably Qualified and Experienced Person (SQEP) is foundational to nuclear operations. Every safety-critical task, every design decision, every inspection and plant intervention depends on the assurance that the responsible engineer possesses demonstrably adequate knowledge, training, and experience — assessed and documented to standards set by the Office for Nuclear Regulation (ONR), the International Atomic Energy Agency (IAEA), and applicable site-level licence conditions.
Yet today, much of this institutional knowledge exists in the minds of a dwindling generation of nuclear engineers — in fragmented documentation systems and in decades of experience that has never been formally encoded. The UK's accelerating decommissioning programme, the growth of Small Modular Reactors (SMRs), and the imperative of NetZero are converging simultaneously, placing unprecedented demands on a nuclear workforce that is already stretched. The question of how to capture, validate, and transfer expert knowledge at scale is not a future problem — it is an operational challenge today.
What Nuclear-SQEP Delivers
Nuclear-SQEP is not a general-purpose large language model applied to nuclear documents. It is a domain-trained model fine-tuned on the specialist vocabulary, reasoning patterns, regulatory frameworks, and engineering methodologies that define expert practice in the UK nuclear sector. The MVP release demonstrates competency across the following domains:
- SQEP framework reasoning — application of SQEP assessment criteria, competency frameworks, and licence condition requirements
- Regulatory intelligence — trained on ONR Safety Assessment Principles, IAEA standards, RCC-M, and UK nuclear legislation
- Decommissioning knowledge — contextual understanding of decommissioning workflows, waste characterisation, ILW container systems, and Sellafield programme considerations
- Thermal hydraulics and reactor physics — foundational nuclear engineering science at SQEP-equivalent depth
- Rotating equipment and electrical machines — grounded in CFD/FEA simulation methodology and structural integrity as applied in nuclear plant operation
- Safety case reasoning — capable of structured safety argument development, hazard identification, and defence-in-depth application
- Knowledge preservation — acts as a scalable, interrogatable repository of nuclear engineering expertise, reducing reliance on individual SQEP availability
Statement from Rotex Technologies
"Nuclear-SQEP represents a landmark moment — not just for Rotex Technologies, but for the entire UK nuclear sector. We have built an AI model that speaks the language of our industry: SQEP, safety cases, decommissioning, regulatory compliance, thermal hydraulics, and rotating equipment. Our mission has always been to electrify towards NetZero, and knowledge infrastructure is as critical to that mission as the physical assets we design and analyse.
This is the world's first open-weight AI model purpose-built for nuclear engineering, and we are releasing it openly because we believe the challenge is too important for any single organisation to address alone. We are now actively seeking investors and technical collaborators who share our conviction that AI-augmented nuclear engineering will be central to delivering the UK's clean energy future. This MVP is the foundation and we are ready to build."
An Invitation to Invest and Collaborate
Nuclear-SQEP is released today as a Minimum Viable Product — a technically credible, domain-grounded foundation that demonstrates the transformative potential of purpose-built AI for the nuclear engineering sector. Rotex Technologies is now actively seeking investors, technology partners, and domain collaborators who wish to participate in the next phase of development, which will encompass:
- Extended domain fine-tuning with curated, proprietary nuclear engineering datasets
- Reinforcement learning from human feedback (RLHF) with SQEP-level nuclear engineers as expert reviewers
- Integration with nuclear document management systems, safety case tools, and engineering simulation platforms
- Deployment solutions for classified and air-gapped nuclear environments
- Validation and verification against ONR Safety Assessment Principles and IAEA standards
- Commercial licensing frameworks for operators, tier-1 contractors, and nuclear training organisations
This is an opportunity to be at the forefront of a genuinely untapped market at the intersection of artificial intelligence, regulated industry, and national energy infrastructure. The nuclear AI space is nascent — Nuclear-SQEP establishes a credible, engineering-first position from which a world-leading capability can be built. Rotex Technologies brings engineering credibility, sector relationships, and technical foundations; we are looking for partners who bring complementary capabilities: compute resource, dataset access, regulatory expertise, or growth capital.
The model is on Hugging Face for research evaluation, benchmarking, and partnership scoping. Commercial licensing and enterprise deployment enquiries are welcomed at info@rotextechnologies.com.
Explore Nuclear-SQEP on Hugging Face Contact Us for Details
About Rotex Technologies
Rotex Technologies Ltd is a UK‑based engineering design and service company specialising in digital twin, industrial machinery,
rotating equipment, AI/ML, manufacturing processes, and sustainable energy solutions. With over 25 years of combined experience,
the team delivers advanced engineering design solutions,
CFD and
FEA simulation,
technical consultancy, and R&D partnerships across nuclear, renewable energy, oil and gas, and advanced manufacturing sectors.
The company is committed to driving digital transformation and supporting the UK's NetZero ambitions.
Website: rotextechnologies.com
About Nuclear-SQEP
Nuclear-SQEP is an open-weight AI language model developed and released by Rotex Technologies Ltd, trained specifically
for the nuclear engineering domain. The model addresses SQEP (Suitably Qualified and Experienced Person) knowledge frameworks,
nuclear regulatory standards, decommissioning engineering, thermal hydraulics, and rotating equipment intelligence.
The MVP release is publicly available on Hugging Face. Commercial licensing and enterprise deployment options are available
upon enquiry.
Model repository: huggingface.co/rotextechnologiesuk/nuclear-sqep
Press & Media Enquiries
For story enquiries, images, investor relations, partnership discussions, or interview requests, please contact us.
Please share it with your network!
